Study on multi-scale-based 3-D surface topography evaluation algorithm

In machining and testing field of MEMS parts, due to anisotropy commonly exists in some technics, the evaluation for 3D surface topography is a difficult program. Existing evaluation algorithm is short of representation capability. Based on discussing of extracting methods of surface topography characteristics, the contourlet Transform (CT) which may provide tight bracing and mutli-scale analysis was introduced. We present a new 3-D surface topography evaluation algorithm. This algorithm has some excellent performances such as multi-scale analysis, high-resolution, time-frequency- localization and multi-directions. And it is good at describing high dimensions data. Meanwhile, according to the theory that noise is not correlated with signal, correlation theory was explored as Discriminant Function. The contourlet transform is rebuilt to an adaptive filter based on this discriminant. Thus high-frequency noise, roughness and waviness information may be separated from. However, we found a question in experiment, so a datum plane was established according to the definition of flatness. Above-mentioned CT and two discriminants may be seen as three low-pass filter banks from the view of filtering. Thus, microcosmical 3-D surface topography of a part may be appraised finely under the three filters work together. The algorithm is verified by simulations. A conclusion can be draw that this algorithm can reliably obtain benchmark of appraisemnt. And the surface information of measured part can be extracted and analyzed without distortion. Comparing with existent 3-D surface topography evaluation algorithm, this method is preponderant in practicality of MEMS and nano surface evaluation.